扫描方式对鲜枣近红外光谱和硬度模型精度的影响  

Influence of Scaning Mode to Fresh Jujube NIR Spectroscopy and Hardness Model Accuracy

作　　者：王斌[1] 薛建新[1] 张淑娟[1] 

机构地区：[1]山西农业大学工学院,山西太谷030801

出　　处：《山西农业科学》2013年第9期934-937,共4页Journal of Shanxi Agricultural Sciences

基　　金：国家自然科学基金项目(31271973);高等学校博士学科点专项科研基金项目(20101403110003);山西省自然科学基金项目(2012011030-3)

摘　　要：以壶瓶枣为研究对象,利用Field Spec 3型近红外光谱仪采集光谱,分析不同扫描方式对壶瓶枣近红外光谱和硬度模型精度的影响。对样品的3个位置分别扫描1,3,6次,采用PLS方法对90个壶瓶枣样品建立硬度模型,其决定系数(R2)均达到0.8以上,校正均方根误差都在0.65以下;应用模型对30个壶瓶枣的硬度进行预测,预测值方差分析结果表明,不同位置进行1,3次扫描所建的硬度模型间无显著差异,确定为较佳扫描方式,而进行6次扫描的硬度模型间有显著差异。在采集光谱建立模型时,应考虑选择合适的扫描次数。Huping jujube as study material, Field Spec3 spectrometer was used for collecting spectra data and analysis of different scaning manners for Huping jujube near infrared spectrum and infuence to hardness model accuracy. On thine positions of the sample, 1 time, 3 times, 6 times were scaned, adopting partial least squares （PLS） regression to establish the hardness model, the coefficient of determination （ R2 ） all reached over 0.8, the calibratio the root mean square error were below 0.65; model was applied to predict the hardness of 30 pot hottles of jujube. Experimental results of three positions 1, 3 times for scanning built by the hardness of the model as the best way to scan, and scaned for six times marked diversity exists in hardness models. When creating a model in the acquisition of spectrum, consideration should be given to choose the appropriate scanning times.

关 键 词：扫描方式 壶瓶枣 近红外光谱 模型精度 方差分析 

分 类 号：S123[农业科学—农业基础科学] S665.1